A Framework Supporting the Development of a Grid Portal for Analysis Based on ROI

Summary

Objectives: In our research on brain function analysis, users require two different simultaneous types of processing: interactive processing to a specific part of data and high-performance batch processing to an entire dataset. The difference between these two types of processing is in whether or not the analysis is for data in the region of interest (ROI). In this study, we propose a Grid portal that has a mechanism to freely assign computing resources to the users on a Grid environment according to the users’ two different types of processing requirements.

Methods: We constructed a Grid portal which integrates interactive processing and batch processing by the following two mechanisms. First, a job steering mechanism controls job execution based on user-tagged priority among organizations with heterogeneous computing resources. Interactive jobs are processed in preference to batch jobs by this mechanism. Second, a priority-based result delivery mechanism that administrates a rank of data significance.

Results: The portal ensures a turn-around time of interactive processing by the priority-based job controlling mechanism, and provides the users with quality of services (QoS) for interactive processing. The users can access the analysis results of interactive jobs in preference to the analysis results of batch jobs. The Grid portal has also achieved high-performance computation of MEG analysis with batch processing on the Grid environment.

Conclusion: The priority-based job controlling mechanism has been realized to freely assign computing resources to the users’ requirements. Furthermore the achievement of high-performance computation contributes greatly to the overall progress of brain science. The portal has thus made it possible for the users to flexibly include the large computational power in what they want to analyze.

Keywords

Grid - portal - development framework - region of interest - MEG

• References

• 1 Mizuno-Matsumoto Y. et al. Propagating process of epileptiform discharges using wavelet-crosscorrelation analysis in MEG. Recent Advances in Biomagnetism. Sendai: Tohoku UniPress. 1999: 782-5.
• 2 Open PBS [computer program]. Version 2.3.16. USA: Altair Engineering, Inc.;
• 3 Frey J. et al. A Computation Management Agent for Multi-Institutional Grids. Cluster Computing 2002; 5: 237-46.
• 4 The Globus Alliance. The Globus Resource Specification Language RSL v1.0. Available at: URL. http://www.globus.org/gram/rsl_spec1.html Accessed 2003
• 5 Czajkowski K. et al. A Resource Management Architecture for Metacomputing Systems. The 4th Workshop on Job Scheduling Strategies for Parallel Processing 1998; 62-82.
• 6 Thomas M, Mock S, Boisseau J. Development of Web Toolkit for Computational Science Portals: The NPACI Hot Page. Proceedings of the 9th IEEE International Symposium on High Performance Distributed Computing 2000; 308-9.
• 7 Novotny J. The Grid Portal Development Kit. Available at: URL. http://www.doescienceGrid.org/Grid/projects/GPDK/gpdkpaper.pdf Accessed 2003